Cache-generated frequently asked questions page

Abstract

In one embodiment, a method for a cache-generated frequently asked questions page includes converting a received query into a set of embeddings and performing a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs. The method further includes returning a stored answer to the received query responsive to a determination that a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair and performing a large language model operation to generate an answer to the query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings.

What is claimed is: 
     
         1 . A method, comprising:
 converting, by a device, a received query into a set of embeddings;   performing, by the device, a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs;   determining, by the device, whether a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings based on whether the received query has a high degree of similarity to a stored embedding of a query in the cache pertaining to the particular question and answer pair;   returning, by the device, a stored answer to the received query responsive to a determination that the particular question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair; and   performing, by the device, a large language model operation to generate an answer to the received query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings.   
     
     
         2 . The method of  claim 1 , further comprising:
 receiving, by the device, user feedback corresponding to the stored answer to the received query or the answer to the received query.   
     
     
         3 . The method of  claim 2 , further comprising:
 logging, by the device, the particular question and answer pair associated with the answer to the received query responsive to receiving the user feedback corresponding to the answer to the received query based on the user feedback being positive or negative.   
     
     
         4 . The method of  claim 1 , further comprising:
 receiving, by the device, a command from a user to perform the large language model operation to generate the answer to the received query subsequent to returning the stored answer to the received query.   
     
     
         5 . The method of  claim 1 , further comprising:
 caching, by the device, the answer to the received query performed by the large language model operation in the cache of question and answer pairs.   
     
     
         6 . The method of  claim 1 , further comprising:
 generating, by the device, a frequently asked questions page using most commonly encountered question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level.   
     
     
         7 . The method of  claim 6 , further comprising:
 providing, by the device, a list containing a particular quantity of most relevant asked questions for the frequently asked questions page.   
     
     
         8 . The method of  claim 1 , further comprising:
 generating, by the device, a hierarchical frequently asked questions page by clustering question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level into topics using unsupervised topic modeling.   
     
     
         9 . The method of  claim 1 , further comprising:
 generating, by the device, a hierarchical frequently asked questions page by clustering question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level based on system administrator input via a user interface.   
     
     
         10 . The method of  claim 1 , further comprising:
 converting the received query into the set of embeddings using a pre-trained sentence transformer, a Bidirectional Encoder Representations from Transformers model that has been fine-tuned for question-answer processing, or a Simple Contrastive Learning of Sentence Embeddings model to create semantically meaningful sentence embeddings, or any combination thereof.   
     
     
         11 . The method as in  claim 1 , further comprising:
 performing the semantic searching operation using an artificial intelligence semantic search technique.   
     
     
         12 . The method as in  claim 1 , wherein performing the large language model operation includes invoking a large language model to compose the answer to the received query that is not present in the cache of the question and answer pairs. 
     
     
         13 . The method as in  claim 1 , further comprising:
 performing the large language model operation using one or more source documents contained within a system associated with the device.   
     
     
         14 . An apparatus, comprising:
 one or more network interfaces to communicate with a network;   a processor coupled to the one or more network interfaces and configured to execute one or more processes; and   a memory configured to store a process that is executable by the processor, the process, when executed, configured to:
 convert a received query into a set of embeddings; 
 perform a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs; 
 determine whether a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings based on whether the received query has a high degree of similarity to a stored embedding of a query in the cache pertaining to the particular question and answer pair; 
 return a stored answer to the received query responsive to a determination that the particular question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair; and 
 perform a large language model operation to generate an answer to the received query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings. 
   
     
     
         15 . The apparatus as in  claim 14 , wherein the process, when executed, is configured to:
 generate a frequently asked questions page using most commonly encountered question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level.   
     
     
         16 . The apparatus as in  claim 14 , wherein the process, when executed, is configured to:
 generate a hierarchical frequently asked questions page by clustering question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level into topics using unsupervised topic modeling.   
     
     
         17 . The apparatus as in  claim 14 , wherein the process, when executed, is configured to:
 generate a hierarchical frequently asked questions page by clustering question and answer pairs in the cache of question and answer pairs that meet the threshold similarity level based on system administrator input via a user interface.   
     
     
         18 . The apparatus as in  claim 14 , wherein the process, when executed, is configured to:
 perform the large language model operation by invoking a large language model to compose the answer to the received query that is not present in the cache of the question and answer pairs.   
     
     
         19 . The apparatus as in  claim 14 , wherein the process, when executed, is configured to:
 performing the large language model operation using one or more source documents contained within a system associated with the apparatus.   
     
     
         20 . A tangible, non-transitory, computer-readable medium storing program instructions that cause a device to execute a process comprising:
 converting a received query into a set of embeddings;   performing a semantic searching operation based on information contained in the set of embeddings against a cache of question and answer pairs;   determining whether a particular question and answer pair in the cache of question and answer pairs meets a threshold similarity level for the information contained in the set of embeddings based on whether the received query has a high degree of similarity to a stored embedding of a query in the cache pertaining to the particular question and answer pair;   returning a stored answer to the received query responsive to a determination that the particular question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings, the stored answer derived from the particular question and answer pair; and   performing a large language model operation to generate an answer to the received query responsive to a determination that no question and answer pair in the cache of question and answer pairs meets the threshold similarity level for the information contained in the set of embeddings.